关于只有一个变点模型的非参数推断  

NONPARAMETRIC METHODS FOR A MODEL WITH AT MOST ONE CHANGE POINT

作　　者：缪柏其[1] 

机构地区：[1]中国科技大学数学系,合肥230026

出　　处：《系统科学与数学》1993年第2期132-140,共9页Journal of Systems Science and Mathematical Sciences

基　　金：国家教委高校博士点基金

摘　　要：这里 μ(t)为非随机函数,μ′(t)为 μ(t)的导函数,α,θ.为未知常数,0<t_0<1,t_0 分别称为模型(1.1)和(1.2)或(1.1)和(1.3)的跳跃变点及斜率变点.θ是模型在变点 t_0处的跃度或斜率跃度,ε(t)为独立随机过程.对每个固定的 t,t∈(0,1),ε(t)的分布与 t 无关.In this paper,the estimates and tests of the change point in a linear regression model arediscussed with Wilcoxon rank sum statistic when observations are large enough.Furthermore,the testing procedure on line is presented.Besides,a method for testing the change point is pro-posed with different observations before and after the change point.

关 键 词：变点模型 统计推断 统计量 

分 类 号：O212.7[理学—概率论与数理统计]